About this course

Most analysts now have a capable AI tool to hand. Using it well is harder. A careless prompt can hand case details to a third party, and a polished AI edit can add facts that were never in your draft.

This course builds on AI Essentials and is hands-on throughout. You'll compare the major platforms and set up a local model for sensitive work, then build prompts, system prompts and a small RAG knowledge base. The second half applies AI in four modes: brainstorming partner, research support, analytical partner and communication support. Each comes with structured techniques, worked examples and guides for citing and logging AI use.

For practitioners ready to move from understanding AI to using it in their own work, with the analyst still in charge.

About 7 hours  |  10 modules  |  Hands-on activities  |  10 quizzes  |  Certificate on completion

What you'll be able to do

Choose the right tool

Match an AI platform or local model to the task and its sensitivity, and know when a public model is off-limits.

Work offline on sensitive cases

Run a local model with LM Studio, GPT4All or Ollama so case material stays on your machine, and give it your own knowledge base.

Collect through AI

Connect a local model to sources such as Ransomware.live, WHOIS and arXiv through MCP, and query them in plain English.

Prompt for intelligence work

Write prompts, system prompts and RAG set-ups that give reliable results for collection, analysis and reporting.

Research faster, verify properly

Use AI to triage articles, read and translate images, geolocate and fill information gaps, then confirm everything against the source.

Challenge and report

Pressure-test assessments with devil's advocate and competing hypotheses, then edit products under a no-new-facts rule and cite your AI use.

What you'll do

1

Decode criminal slang

Work out what terms like "Fullz" and Telegram emoji codes mean, then build a keyword list and check every term before you search.

2

Read a suspect company's sign

Extract and translate the sign at a company being checked for links to North Korea, and compare what different models give you.

3

Recover a hidden email

Generate likely candidates for a partially redacted email address, then confirm them with email verification tools.

4

Check for state ownership

Test whether a Chinese company is state-owned, and pull names, roles and links from a foreign-language annual report.

5

Catch mistranslations

Compare AI translations of Russian or Chinese statements at the UN Security Council with the official record.

6

Challenge an assessment

Run AI as devil's advocate on an assessment of Russian activity against subsea infrastructure, and decide which challenges stand up.

What's included

Ten modules, about 45 minutes of video, ten knowledge-check quizzes and hands-on activities throughout. You also keep five best practice guides: Managing AI Integration in OSINT, A Framework for Evaluation of AI Outputs, Citing AI Used in OSINT Workflows, Making Smart Selections in Training and Tooling, and Secure & Resilient: Setting OSINT Teams Up for Success.

Time, audience & prerequisites

Estimated time to complete: about 7 hours, including reading, videos, quizzes and activities. Allow extra time if you set up a local model as you go. This course is self-paced: complete it in a few sittings or come back whenever it suits you over 12 months.

Who this course is for: OSINT practitioners, analysts and investigators who understand how generative AI works and now want to use it in their planning, collection, analysis and reporting.

What you'll need: a web browser and access to at least one AI chat tool, such as ChatGPT, Claude or Gemini. Paid tiers give better results but aren't required. The local model lessons need a computer you're allowed to install software on, and the larger setups use Docker. If your organisation restricts AI use, you can still follow the demonstrations.

Prerequisites: AI Essentials, or an equivalent understanding of how generative AI works, including hallucination and sycophancy. The first quiz checks this. Foundational OSINT experience, such as OSINT Primer, will help.

What's next?

Take your AI work further with Agentic AI for OSINT, or put it to use in a specialist area.

Location and Event Intelligence with Open Sources →

Applies AI-assisted geolocation, monitoring and regional assessments to places and events.

Searching the Chinese Internet →

Puts AI translation and prompting to work across Chinese-language search engines, platforms and sources.

Get all three AI courses in the AI Bundle

View the bundle

COURSE CURRICULUM

    1. Welcome

    2. Before We Begin

    3. The Three Reference Guides

    4. Applying the AI Modes

    5. Failure Mode Focus: OPSEC

    6. Quiz: Leveraging Gen AI for OSINT

    1. Working with AI Platforms

    2. ChatGPT (OpenAI)

    3. Claude (Anthropic)

    4. Grok (X)

    5. Gemini (Google)

    6. DeepSeek

    7. Quiz: From Understanding to Application

    1. Embedded Models

    2. Google AI Mode

    3. Copilot (Microsoft)

    4. Meta AI

    5. Brave AI

    6. Image and Video Generation Tools

    7. Choosing the Right Tool for the Task

    8. Quiz: Embedded and Use-Case Specific AI

    1. AI Transparency Statement

    2. Offline and Local Models

    3. Pros and Cons of Running a Local Model

    4. Guardrails and OSINT Challenges

    5. Local Models: Use Cases for OSINT

    6. Local Model Setup

    7. Local Models and Local Knowledge Bases

    8. Leveraging Markdown in Local Models

    9. Activity: Markdown

    10. Model Context Protocol

    11. MCP Use Cases and Setup

    12. Quiz: Local and Offline AI Models

    1. What is Prompt Engineering

    2. Goals, Context and Constraints

    3. Activity: Applying Goals, Context and Constraints

    4. Clarity

    5. Prompting With Examples (Few-Shot Prompting)

    6. Chain-of-Thought (CoT) Prompting

    7. Using XML Tags

    8. Chunking

    9. Engineering Prompts with AI

    10. Prompting Tips for Reasoning Models

    11. Quiz: Prompt Engineering

    1. From Prompt Engineering to Context Engineering

    2. System Prompts and Custom Instructions

    3. Retrieval-Augmented Generation (RAG)

    4. Custom GPTs and Claude Projects

    5. Activity: Set Up a Basic RAG Pipeline for a Document Collection

    6. Quiz: Context Engineering and RAG

COURSE SNAPSHOT

  • $399.00
  • 85 lessons